Assessment of Interlaboratory Performance on the Measurement of Blood Lead Levels in Taiwanese Adults

Transcription

1 , f Industrial Health, 1995, 33, Assessment of Interlaboratory Performance on the Measurement of Blood Lead Levels in Taiwanese Adults Saou-Hsing LIOU1), Guang-Yang YANG2), Trong-Neng WU3) Ying-Chin K04~, Ching-Chang LEES, Shien-Tsong 1106), Jim-Shoeing LAI'~, Yea-Quay WU8~, Horn-Che CHIANG3~, Kquei-Nu K03~ and Po-Ya CHANG3~ F. 1) School of Public Health, National Defense Medical Center, Taipei 2) Division of Clinical Toxicology, Department of Medicine, Taipei Veterans General Hospital, Taipei 3) Department of Health, The Executive Yuan, Taipei 4) School of Public Health, Kaohsiung Medical College, Kaohsiung 5) Department of Occupational and Environmental Health, National Cheng Kung University Medical College, Tainan 6) Department of Industrial Safety and Hygiene, Chia-Nan Jr. College of Pharmacy, Tainan 7) Institute of Environmental Health, China Medical College, Taichung 8) Department of Family and Occupational Medicine, Taiwan Provincial Hsinchu Hospital, Hsinchu (Received July 4, 1995 and in revised form October 6, 1995) Abstract: The purpose of this proficiency study was to assess the accuracy and consistency of blood lead level (BLLs) measurements in a study of Taiwanese adults. Three methods, including a certified blood accuracy test, an interlaboratory precision test and an intertime repeated measurement ;test, were applied to the six participating laboratories. Accuracy tests showed that most of the blood lead measurements were within the acceptable criteria proposed by United States Center for Disease Control and Prevention (U.S. CDC). However, an average of 11% underestimation was found at BLLs below 15 pg/dl. Coefficients of variation (CVs) were high in the first 3 months. After technical improvements9 the CVs were reduced to acceptable limits of around 15% at low target lead values and 7 % at high lead values. Interlaboratory variations of measurements in blood from ten normal healthy donors showed that the standard deviations were less than 2 µg/dl, which is within the acceptable criteria of ±4 µg/dl, in 8 out of 10 samples. Correspondent: Dr. Saou-Hsing Liou, P.O. Box , School of Public Health, National Defense Medical Center, Taipei, Taiwan, 10107, R.O.C. Telephone: FAX:

2 182 S.H. LIOU et al. Repeated measurements of BLLs in 54 blood samples over two months showed that most of the differences were within the acceptable range with a few exceptions. The mean BLLs measured in the baseline (pre-test) and two months later (post-test) were nearly identical. This proficiency test provides comparable and reliable results of BLL estimations in this multilaboratory study. However, the accuracy and consistency at low BLLs need to be improved. Key words: Blood lead level - Accuracy test - Precision test - Interlaboratory variations - Intertime variations INTRODUCTION Although a blood lead level is a good biomarker of lead absorption, variations of blood lead measurements exist within and between laboratories1-10~. The variations in measurement of blood lead levels (BLLs) may derive from random and systematic errors. A proficiency test program is applied to remove systematic error and to reduce the random error of measurements, especially in a multilaboratory study. Requirements for accurate measurements of blood lead concentrations become more stringent as the baseline BLL of the general population declines. Cases of lead poisoning due to occupational lead exposure and environmental pollution have been documented for more than 10 years in Taiwan". Measurement of blood lead concentrations was first introduced into Taiwan in the early 1980's, but a program to evaluate accuracy and consistency among laboratories was still lacking. A study was designed to measure the BLLs of Taiwanese adults in order to evaluate lead exposure through environmental sources12~. The validity and reliability of blood lead measurements of the participating laboratories were monitored in order to obtain accurate estimations of BLLs. This paper presents the performance of the six participating laboratories with comparison to the criteria for proficiency tests proposed by the U.S. Center for Disease Control and Prevention (CDC). Recently, a surveillance program was initiated in Taiwan to monitor occupational and environmental lead exposure in the population'3' 14) This study provides a basis of quality control toward successful progress in the establishment of a reporting system. MATERIALS AND METHODS A total of 3,000 blood samples, collected from a random sample of Taiwanese adults, were distributed equally to the six participating laboratories for blood lead measurements. These six laboratories are located in medical centers or medical schools where blood lead measurements have been routinely performed for several years. All of the laboratories use flameless atomic absorption spectrophotom-

3 ASSESSMENT OF BLOOD LEAD MEASUREMENTS IN TAIWAN 183 etry (AAS) and have similar protocols for blood lead measurements. The proficiency test was conducted and supervised by a medical center laboratory which did not participate in the sample analysis. This supervising laboratory had also participated in proficiency tests provided by the Quebec Center for Toxicology Interlaboratory Comparison program, Canada and the U.S. CDC. The laboratory technicians were trained in a workshop on blood lead measurement held three months before this study. An exercise of this proficiency test program was held from June to August. The formal running of this proficiency test began prior to (September to December) and continued through the period in which measurements of blood samples from Taiwanese adults were performed (January to May of the next year). A laboratory audit was held at the end of November in order to improve the internal quality control and the consistency of procedures in blood lead analysis. Three quality control methods were used in this proficiency test program. First, certified whole blood controls (Kaulson Laboratories, Inc., New York), which included 11 blood lead concentrations ranging from 10 to 100 µg/dl, were selected for the accuracy test. Three randomly selected blood controls including low (10 to 22 µg/dl), medium (38 to 55 µg/dl), and high (79 to 100 µg/dl) target lead levels were mailed blindly to these 6 laboratories monthly from September to May. In order to identify the sample deterioration that may have occurred in the mailing process, the supervising laboratory concurrently measured the certified whole blood samples. The results of the measurements, including the laboratory records, readings of three tested blood samples and intralaboratory variations, were sent back to the supervising laboratory within two weeks. Any problem with an AAS machine or any change of technicians in this period was reported to the supervising laboratory. Second, 10 blood samples from normal healthy donors were distributed blindly to the six laboratories for testing of interlaboratory variations during measurements of BLLs (March). None of these 10 donors had a known history of occupational lead exposure. All donors were nonsmokers with the exception of one individual who smoked 20 cigarettes per day for 20 years. Third, 54 blood samples (nine samples from each laboratory) selected from the study population were blindly and repeatedly measured over two months (March and May) by the six laboratories to evaluate the intertime variations and intralaboratory consistency of blood lead measurements. All of these whole blood samples were kept at -20 C. The pilot study showed that BLLs do not change over six months if blood samples are kept frozen at -20 C. Statistical methods The blood lead levels are presented as mean and standard deviations. Coefficients of variation (CVs) are used to present the variations among laboratories,

4 184 S.H. LIOU et al. time and testings. Paired t-test was used for comparison of repeated measurements. RESULTS Accuracy test of certified whole blood Three blood samples at high, medium and low target lead values of certified whole blood controls were sent to these six laboratories as well as the supervising laboratory monthly. The mean BLLs measured by the six laboratories with comparison to the 11 certified target values are shown in Table 1. The mean BLLs measured by the six laboratories were nearly within the criteria proposed by the CDC, U.S. Public Health Service6~, that is, ±4 µg/dl if blood lead value < 40 µg/dl and ±10% if value >_ 40 µg/dl. Although there was no systematic trend with time, most of the laboratories underestimated the true target values. Since most of the BLLs in the nonexposed adults were less than 15 µg/dl, the certified blood controls at 10, 12, and 15 µg/dl target lead values were used to anaiyze the accuracy of the blood lead measurements among the six laboratories (Table 2). Three blood controls at the target lead value of 10 µg/dl were measured at Fa mean of 9.2 µg/dl, four controls at 12 µg/dl were measured at a mean of 10.3 ` µg/dl, and four controls at 15 µg/dl were measured at a mean of 13.4 µg/ dl. The underestimations were 8%, 14% and 11%, respectively, at these three target lead values. The average of the underestimations was 11%. These results showed that there was an underestimation of 11% at BLLs less than 15 µg/dl. The underestimatian was 8% and 11% at the target values of 100 µg/dl and 95 µg/ dl, respectively (Table 1). Table 1. Results of accuracy 'tests: means and CVs (%) of 6 laboratories compared to target values.

5 ASSESSMENT OF BLOOD LEAD MEASUREMENTS IN TAIWAN 185 Table 2. Accuracy of laboratories. estimations at blood lead levels below 15 µg/dl in the 6 Consistency of measurements between laboratories was expressed as coefficients of variation (CVs). The CVs among 6 laboratories of the 11 target lead values are shown in Table 1. The CVs in the first three months (September to November) were high, but declined to acceptable levels in the last 6 months. In the first 3 months, the CV for the low target value was as high as 37%, and the CVs for the medium and high target values were approximately 15%. A laboratory audit was held to improve the consensus of techniques. In the last 6 months, the CVs decreased to between 12% and 22% at the low target values (10 to 22 µg/dl), between 9% and 16% at medium target values (38 to 55 µg/dl), and 7% at the high target values (100 µg/dl). In general, the CVs were higher at the low target lead values than at the medium and high values. However, the interlaboratory CVs in the last 6 months are considered to be acceptable with comparison to the CVs of other studies2, 6, s). Precision test of interlaboratory variations Blood samples drawn from ten normal healthy donors without known history of lead exposure were sent blindly to the six laboratories in order to evaluate the interlaboratory variations. The mean and standard deviations of blood lead levels measured by the six laboratories are shown in Table 3. During this precision test, laboratory No. 3 reported that their AAS instrument was out of order and their results were excluded in the data analysis. The standard deviations of the other five laboratories were less than 2µg/dl in 8 of 10 donors. Two donors were outside this range at 2.2 and 3.3 µg/dl. A standard deviation of less than 2 µg/ dl indicates that the interlaboratory variations meet the CDC criteria of ±4 µg/dl, which is equivalent to 2 standard deviations. In addition, there was no statistical difference of mean BLLs measured by these five laboratories. The only smoking donor (No. 3) showed the highest BLL in this study population, which was compatible with findings from our study and others12~.

6 186 S.H. LIOU et al. Table 3. Interlaboratory variations of blood lead levels of 10 individuals. Precision test of intertime repeat measurements Fifty-four blood samples selected from the study population and measured by the six laboratories were blindly and repeatedly measured over two months in order to evaluate the intralaboratory and intertime variations. The results of the baseline (pre-test) and repeat measurements of BLLs two months later (post-test) are shown in Table 4. Eleven out of 54 blood samples showed discrepancies in the values of the repeat measurements. However, most of the differences between the pretest and post-test levels were within the acceptable criteria of ±4 µg/dl. The mean BLL of the pre-test was sg/dl, which was nearly the same as the mean BLL of the post-test (11.01 sg/dl). The mean difference between the pre-test and the post-test (0.06 µg/dl) was not statistically different by the paired t-test. These results reveal that variations exist in the repeat measurements, however, the means of the measurements are consistent over a two month period. The results imply that the time variations of the blood lead measurements may not affect the estimation of the mean BLL in Taiwanese adults. DISCUSSION Laboratory performance of blood lead measurements has been assessed in a number of interlaboratory studies over the last 20 years1-10~. In Taiwan, this is the first interlaboratory proficiency study since the first introduction of atomic absorption spectrophotometry in the early 1980's. This proficiency test was conducted to determine the accuracy and consistency of blood lead measurements among

7 ASSESSMENT OF BLOOD LEAD MEASUREMENTS IN TAIWAN 187 Table 4. Consistency of blood lead measurements over two months.

8 188 S.H. LIOU et al. laboratories in a study of the BLLs in Taiwanese adults. The six laboratories that participated in this study were either in medical centers with an occupational medicine clinic or in medical schools with a department of industrial hygiene. These laboratories routinely provide blood lead measurements for clinical service or research. Accurate measurements of. BLLs are more stringent since the baseline BLLs of the general population are declining15-21). Most of the measurements of certified whole blood controls and the means of the six laboratories were within the acceptable limits proposed by U.S. CDC6~. Before July 1, 1990, performance was considered satisfactory if BLLs were measured within ±6 µg/dl of the target value at lead levels < 40 µg/dl, or within ±15% at lead values >_ 40 µg/dl. After July 1, 1990, the acceptable criteria were tightened to ±4 µg/dl for lead values < 40 µg/dl and ±10% for lead values >_ 40 µg/dl6~. Most of the results in this proficiency study meet these new criteria. In this study, the means of the BLLs tended to be underestimated at both low and high target lead values. The underestimation of BLLs was about 11% in this study. This observation conflicted with the results from most of the other interlaboratory studies. Usually, laboratories tended to overestimate at BLLs below 40 µg/dl and underestimate at BLLs above 40 µg/dl's. However, underestimates at BLLs below 40 µg/dl and overestimates at BLLs above 40 µg/dl have also been reported6' 7) These results suggest that underestimation of BLLs in Taiwan may have occurred in the past 10 years. The reason for the underestimation may be due to the fact that most laboratories in Taiwan used only one commercially available reference standard or proficiency test samples from the U.S. CDC to establish the internal standard curve. It has been strongly suggested that proficiency test samples may not be used for internal quality control because of instability in long-term use6~. Commercial reference standards may also not be stable enough for long-term storage. Reliable reference standards may be purchased from other reputable sources to minimize such underestimation. Proficiency testing of the type described in this study was found to improve the consistency of blood lead measurements through suitable training and rigorous monitoring. The variations of blood lead measurements among the six laboratories were significant in the first three months. Thereafter, the CVs decreased to acceptable levels when the laboratory technicians were trained and an audit was held. The CVs in this study are about 12% to 22% at blood lead levels below 22 µg/dl, and 9% to 16% for blood levels between 38 and 55 µg/dl. This performance is comparable to that noted in other studies2' 6, 8) which described interlaboratory variations of 10% to 22% at blood lead levels approximately 50 µg/dl. Interlaboratory consistency in measurements of blood samples from normal healthy donors showed that eight out of ten samples had standard deviations below 2 µg/dl. This means that the variations were within the acceptable criteria of ±4

9 ASSESSMENT OF BLOOD LEAD MEASUREMENTS IN TAIWAN 189.tg/dl. These results show consistency of blood lead measurements among laboratories, which reduces interlaboratory bias in the estimation of BLLs of Taiwanese adults. The results of repeated measurements over two months also provide evidence of intralaboratory and intertime consistency. This precision test helps exclude the time variations in blood lead measurements, which becomes important when blood samples are collected over a period of time. It is also important to evaluate the time trend of BLLs. This proficiency test program created a model for multicenter study of BLLs in Taiwan. In addition to accuracy, this proficiency test also evaluated the intralaboratory, interlaboratory and intertime consistency of BLL measurements. This program will be continued and extended to 22 laboratories legally capable of measuring blood lead levels in Taiwan in order to improve the performance of blood lead measurements and provide basic quality control for the surveillance system of increased BLLs. ACKNOWLEDGMENTS This study was supported by a grant from Department of Health, The Executive Yuan, R.O.C. [DOH8I-HP-201]. The authors thank David Jacobson-Kram, Ph.D. for the manuscript preparation. REFERENCES 1) Boone J, Hearn T, Lewis S. Comparison of interlaboratory results for blood lead with results from a definitive method. Clin Chem 1979; 25: ) Bullock DG, Smith NJ, Whitehead TP. External quality assessment of assays of lead in blood. Clin Chem 1989; 32: ) Claeys F, Ducoffre G, Sartor F, Roels H. Analytic quality control of cadmium and lead in blood and cadmium in urine: results of its implementation during a five-year epidemiological study. In "Cadmium in the Human Environment: Toxicity and Carcinogenicity" (Nordberg GF, Herber RFM, and Alessio L, Eds), Lyon, International Agency for Research on Cancer, IARC, 1992; p ) Donovan DT, Vought VM, Rakow AB. Laboratories which conduct lead analyses on biological specimens. Arch Environ Health 1971; 23: ) Keppler JF, Maxfield ME, Moss WD, Tietjen G, Linch AL. Inter-laboratory evaluation of the reliability of blood lead analyses. Am Ind Hyg Assoc J 1970; 31: ) Parson PJ. Monitoring human exposure to lead: an assessment of current laboratory performance for the determination of blood lead. Environ Res 1992; 52: ) Rollin HB, Kilroe-Smith TA, Theodorou P. Quality control of analyzing lead in blood: Evaluation and comparison of participating laboratories. S Afr J Sci 1988; 84: ) Saltzman BE. Variability and bias in the analyses of industrial hygiene samples. Am Ind Hyg Assoc J 1985; 46: ) Subramanian KS. Determination of lead in blood-an interlaboratory study. Sci Total Environ 1988; 71: ) Wu YQ, Tang XY, Li XL, Lu YF, Tang J. Quality assurance in the biological monitoring of lead exposure in China. Int Arch Occup Environ Health 1993; 65: s231-4.

10 190 S.H. LIOU et al. 11) Liou SH, Gu TL, Hsu SW, Wu DM, Chen LM. A review of occupational and environmental lead poisoning in Taiwan. J Occup Safety Health (ROC) 1994; l: ) Liou SH, Wu TN, Chiang HC, Yang GY, Wu YQ, Lai JS, et al. Blood lead levels in the general population of Taiwan, Republic of China. Int Arch Occup Environ Health 1994; 66: ) Wu TN, Shen CY, Yang GY, Liou SH, Ko KN, Chao SL, Hsu CC, Lai JS, Chang PY. Occupational lead surveillance-taiwan, July-December MMWR 1995; 44: 181, ) Wu TN, Shen CY, Yang GY, Liou SH, Ko KN, Chiang HC, Lai JS, Ho CK, Chang PY. Establishment of an occupational disease surveillance system to monitor blood lead levels in Taiwan. Prey Med 1995; 24: ) Brody DJ, Pirkle JL, Kramer RA, Flegal KM, Matte TD, Gunter EW, et al. Blood lead levels in the US population. Phase 1 of the third National Health and Nutrition Examination Survey (NHANES III, 1988 to 1991). JAMA 1994; 272: ) Ducofre G, Claeys F, Bruaux P. Lowering time trend of blood lead levels in Belgium since Environ Res 1990; 51: ) Elinder CG, Friberg L, Lind B, Nilsson B, Svartengren M, Oevermark I. Decreased blood lead levels in residents of Stockholm for the period Scand J Work Environ Health 1986; 12: ) Hinton D, Coope PA, Malpress WA, Janus ED. Trends in blood lead levels in Christchurch (NZ) and environs J Epidemiol Community Health 1986; 40: ) Pirkle JL, Brody DJ, Gunter EW, Kramer RA, Paschal DC, Flegal KM, et al. The decline in blood lead levels in the United States. The National Health and Nutrition Examination Surveys (NHANES). JAMA 1994; 272: ) Quinn MJ, Delves HT. UK blood lead monitoring programme : Protocol and results for Hum Toxicol 1987; 6: ) Quinn MJ, Delves HT. UK blood lead monitoring programme : Results for Hum Toxicol 1988; 7: